Farmer School team wins 2020 Teradata Analytics Challenge
Oct 01, 2020
A team of Farmer School students took on students from around the world to win the analytics portion of the 2020 Teradata Analytics and Data Challenge last month. Claire Galberg, Yili (Toby) Yang, Brant Imhoff, and Laura de Oliveira competed against seven other finalist teams, including two from Miami, to win the challenge.
The path to winning the challenge started last semester. “We submitted a condensed version of a project we worked on in Dr. Jay Shan’s ISA 514 course (managing big data). We were challenged by Dr. Shan to work on any analysis we chose using any type of tools or models we wanted,” Galberg explained. “We decided to analyze COVID-19 cases because we began the project in the middle of last semester, when there were a lot of unknowns and a big media frenzy. We tried to build a regression model to predict COVID-19 deaths based on other variables like population, and a time series analysis of the number of cases and deaths to predict what those numbers would be for a 2-week period.”
“We built two time series models for log deaths and log cases for the US, trained on data from January through April. Both were Seasonal ARIMA models that were chosen based on cross-validation error rates, then transformed to predict the first couple weeks' cases and deaths in May,” Imhoff said. “Our models suggested a jump in U.S. cases, coinciding to increased testing availability and a temporary decline in deaths. We suggested then that it might not be quite the time to reopen and authorities should proceed with caution, assuming a decline in cases for 14 days was the precedent.”
Galberg and de Oliveira worked to prepare the study for the competition, as Imhoff and Yang had graduated in May. “We kept our team motivation up and continually pushed the project forward. We had good communication within the team, which kept everyone being productive,” Yang remarked. “We were still communicating while finalizing the presentation even though I was 6 hours ahead in Switzerland. I am grateful and lucky for having teammates like them.”
After being named finalists in August, the selected teams made virtual final presentations in mid-September, then learned they had won the Analytics portion of the challenge. “The competition was a great opportunity for us to be exposed to students from other schools and see the tools they used and the analyses they performed,” Galberg said. “I’m beginning to learn there’s really no limit to what you can do with data. It’s not all programming and math, there’s ambiguity and room for creativity. I’m really grateful for what the faculty in the ISA department has taught me, and for opportunities like this to learn from other students as well.”
The other Miami teams that made the analytics finals were the team of Amy Hu, Yao Luo, and Ethan Rosser, and the team of Helen Zhang and Alex Kwon.
"I am very proud with the performance of all three Miami teams who were selected as finalists in this prestigious competition, with a special congratulations to the team finishing on top. Our teams tackled a diverse set of interesting, important, and globally relevant issues,” Shan said. “All of our teams brought forth great passion and creativity along with their strong analytical skills. It is another testament to the experiential learning in our curriculum."
The path to winning the challenge started last semester. “We submitted a condensed version of a project we worked on in Dr. Jay Shan’s ISA 514 course (managing big data). We were challenged by Dr. Shan to work on any analysis we chose using any type of tools or models we wanted,” Galberg explained. “We decided to analyze COVID-19 cases because we began the project in the middle of last semester, when there were a lot of unknowns and a big media frenzy. We tried to build a regression model to predict COVID-19 deaths based on other variables like population, and a time series analysis of the number of cases and deaths to predict what those numbers would be for a 2-week period.”
“We built two time series models for log deaths and log cases for the US, trained on data from January through April. Both were Seasonal ARIMA models that were chosen based on cross-validation error rates, then transformed to predict the first couple weeks' cases and deaths in May,” Imhoff said. “Our models suggested a jump in U.S. cases, coinciding to increased testing availability and a temporary decline in deaths. We suggested then that it might not be quite the time to reopen and authorities should proceed with caution, assuming a decline in cases for 14 days was the precedent.”
Galberg and de Oliveira worked to prepare the study for the competition, as Imhoff and Yang had graduated in May. “We kept our team motivation up and continually pushed the project forward. We had good communication within the team, which kept everyone being productive,” Yang remarked. “We were still communicating while finalizing the presentation even though I was 6 hours ahead in Switzerland. I am grateful and lucky for having teammates like them.”
After being named finalists in August, the selected teams made virtual final presentations in mid-September, then learned they had won the Analytics portion of the challenge. “The competition was a great opportunity for us to be exposed to students from other schools and see the tools they used and the analyses they performed,” Galberg said. “I’m beginning to learn there’s really no limit to what you can do with data. It’s not all programming and math, there’s ambiguity and room for creativity. I’m really grateful for what the faculty in the ISA department has taught me, and for opportunities like this to learn from other students as well.”
The other Miami teams that made the analytics finals were the team of Amy Hu, Yao Luo, and Ethan Rosser, and the team of Helen Zhang and Alex Kwon.
"I am very proud with the performance of all three Miami teams who were selected as finalists in this prestigious competition, with a special congratulations to the team finishing on top. Our teams tackled a diverse set of interesting, important, and globally relevant issues,” Shan said. “All of our teams brought forth great passion and creativity along with their strong analytical skills. It is another testament to the experiential learning in our curriculum."